Exchange rates play a fundamental role in determining purchasing power and relative wealth between countries. Since the severing of the Bretton Woods system of fixed exchange rates in the 1970s, most advanced countries opted to let their exchange rates float according to prevailing market forces. Since then, it has been surprisingly difficult to find economic fundamentals or other variables that can explain exchange rates in accordance with classic models (Meese and Rogoff 1983), a phenomenon dubbed the ‘exchange rate disconnect’. This continuing difficulty is a source of discomfort for policymakers and frustration for researchers, as summarised by Rossi (2013). The ability to explain exchange rates remains a critical item on the international macroeconomics research agenda.
A reconnect after the Global Crisis
It is against this backdrop that we uncover a surprising pattern that emerged with the Global Crisis. Exchange rates, and in particular the broad US dollar, have co-moved closely with global risk premia and with US foreign bond purchases. Since 2007, when proxies for global risk premia increase (i.e. when the prices of risky assets broadly decline), the dollar contemporaneously appreciates. Whereas risk measures had little or no explanatory power for exchange rates prior to the crisis, these same risk measures statistically explain a meaningful share of all subsequent exchange rate variation. Similarly, when the US is a net purchaser of foreign bonds, the dollar depreciates. We dub the emergence of the relationships of global risk proxies and US foreign bond purchases with the exchange rate an ‘exchange rate reconnect’ (Lilley et al. 2019).
The reconnect of exchange rates with global risk proxies can be clearly seen in Figure 1, which plots the R2 values of rolling univariate regressions of the monthly change in the broad dollar exchange rate on the contemporaneous change in six global risk proxies. These proxies include (i) the ‘GZ Spread’, an index of aggregated US corporate bond spreads constructed by Gilchrist and Zakrajšek (2012); (ii) the ‘VXO’, calculated as the monthly change in the log implied volatility on the S&P100 stock index; (iii) the log total return on the S&P500; (iv) the ‘Treasury Premium’ constructed as the average one-year covered interest parity deviation between developed country government bonds and US Treasuries taken from Du et al. (2018); (v) the ‘Global Factor’, the common component of movements in world asset prices constructed by Miranda-Agrippino and Rey (2018); and (vi) the “Intermediary Returns” from a value-weighted portfolio of holding companies of New York Federal Reserve primary dealers taken from He et al. (2017). The top panel shows R2 measures from regressions on 10-year (i.e. 120-month) rolling windows while the bottom panel shows them for 5-year rolling windows.
Figure 1 The US dollar and measures of risk premia
Note: Figure plots the 120- and 60- month rolling R2 for regressions of the average log change in the US dollar versus the other G10 currencies against various indicators of risk. Both panels use data samples beginning in January of 1977 and ending in December of 2018.
Around 2007 there is a sharp and sustained change in the explanatory power of all of these risk proxies for the broad US dollar, as shown in Figure 1. Prior to 2007, almost all of these risk measures have an R2 below 5%. After 2007, all measures have an R2 for the regressions using 5-year windows that peaks at 40% or higher. As can be seen, in the most recent data, the explanatory power of these variables has significantly declined. Though most of the measures end our sample with R2 values above their pre-2007 highs, the recent data beg the question of whether this reconnect is a post-crisis phenomenon or a permanent structural break.
In addition to documenting the reconnect with price-based measures, we show that during 2007-2012 an aggregate financial quantity, US purchases of foreign bonds, reconnected both with the measures of global risk appetite and with the broad US dollar. In quarters when US residents increased their holdings of external debt, the dollar contemporaneously depreciated. Figure 2 shows the corresponding rolling R2s.
Figure 2 The US dollar and US foreign bond purchases
Note: In the top panel, the figure plots the 20- and 40-quarter rolling R2 of a regression of the log appreciation of the US dollar against all other G10 currencies on US net purchases of foreign bonds, normalised as a percentage of the US value of foreign bond investment at the end of the prior quarter. In the bottom panel, the y-axis corresponds to the quarterly change in the US dollar against all other G10 currencies, defined such that a positive value corresponds to a depreciation. The x-axis shows the purchases of foreign bonds by the US in the same quarter. Regression lines are estimated using the full sample (2007:Q1 to 2019:Q2) and excluding the crisis (2009:Q3 to 2019:Q2).
The relationship between the broad dollar and US foreign bond purchases appears to have disappeared in the most recent period of 2013-2018. This raises the question of why currencies began to strongly covary with measures of global risk and these flows at the time of the global financial crises. We use rich micro-level security holdings data from Morningstar assembled in Maggiori et al. (2019, 2020), to examine this question, and we find that at the same time that US residents were buying foreign securities, they were also selling US treasuries. This behaviour is consistent with the idea that the US provides a key role in the global financial system – providing safe assets to foreign countries and in turn taking a share of their risk (Caballero et al. 2008, Gourinchas et al. 2011, Maggiori 2017, Farhi and Maggiori 2018).
In the context of the voluminous literature on exchange rate disconnect, which offers few comparably successful covariates, we consider this to be progress even if the post-crisis time series is short. Going forward, anyone analysing the relationship between exchange rates and risk should pay close attention to the large structural break that we document over the last 10 years.
Authors’ note: For more information and the codes and data behind this column, see the Global Capital Allocation Project.
Caballero, R J, E Farhi, and P-O Gourinchas (2008), “An Equilibrium Model of "Global Imbalances" and Low Interest Rates”, American Economic Review 98(1): 358–393.
Coppola, A, M Maggiori, B Neiman, and J Schreger (2019), “Redrawing the Map of Global Capital Flows: The Role of Cross-Border Financing and Tax Havens”, Working Paper.
Du, W, J Im, and J Schreger (2018), “The U.S. Treasury Premium”, Journal of International Economics 112: 167–181.
Farhi, E, and M Maggiori (2018), “A Model of the International Monetary System,” The Quarterly Journal of Economics 133(1): 295–355.
He, Z, B Kelly, and A Manela (2017), “Intermediary asset pricing: New evidence from many asset classes”, Journal of Financial Economics 126: 1–35.
Gilchrist, S, and E Zakrajšek (2012), “Credit spreads and business cycle fluctuations,” American Economic Review 102(4): 1692–1720.
Gourinchas, P-O, N Govillot, and H Rey (2011), “Exorbitant Privilege and Exorbitant Duty”, Working Paper UC Berkeley and LBS.
Lilley, A, M Maggiori, B Neiman, and J Schreger (2019), “Exchange Rate Reconnect”, Working Paper.
Maggiori, M (2017), “Financial Intermediation, International Risk Sharing, and Reserve Currencies” American Economic Review 107(10): 3038–3071.
Maggiori, M, B Neiman, and J Schreger (2020), “International Currencies and Capital Allocation”, Journal of Political Economy, forthcoming.
Maggiori, M, B Neiman, and J Schreger (2019), “The Rise of the Dollar and Fall of the Euro as International Currencies”, AEA Papers and Proceedings 109: 521-26.
Meese, R A, and K Rogoff (1983), “Empirical exchange rate models of the seventies: Do they fit out of sample?”, Journal of International Economics 14(1-2): 3–24.
Miranda-Agrippino, S, and H Rey (2019), “US Monetary Policy and the Global Financial Cycle”, NBER Working Paper no. 21722.
Rossi, B (2013), “Exchange rate predictability”, Journal of Economic Literature 51(4): 1063–1119.